Replacing or repairing TR-808 switches and pots
2012-08-31 Robin Whittle
"Death and taxes are unsolved engineering problems" - Romana Machado.
This page discusses the repair or
replacement of all the TR-808's
switches and pots, with a final detailed section on the Run/Stop
keyswitch. There is an unsolved engineering problem regarding
these switches. They fail and are not easily replaced or
repaired. Please let me know your experiences or any suggestions
- I can link to other pages or add your ideas to this page.
You may also be interested in my analysis of tact switches for the
The TR-808 is a mains-operated
device with exposed mains connections
on the power switch and the power supply board. Working on this
machine while it is plugged into a mains socket constitutes a serious
safety hazard. To so so risks serious
injury and DEATH. If you are not an experienced electronic
technician who already knows about mains safety and about static
electricity damage to semiconductor devices, I suggest you not
dismantle this machine at all.
The Volume pot
may become noisy. My fix for this is to de-solder it, dismantle
by bending tabs, and then to clean the contacts with isopropyl alcohol.
I usually spin it around a few times (be sure to do so in the direction
where the wipers are dragged, rather than pushed) and use little strips
of cardboard soaked in
isopropyl alcohol. As far as I know, this fixes the problem for a
few years at least.
Technology Transplant http://www.technologytransplant.com/
sell newly manufactured Volume and Tempo
pots. I haven't seen any trouble with Tempo pots, but they too
could become erratic and benefit from cleaning with isopropyl
alcohol. Likewise the Fine Tempo
Generally, I advise never spraying anything anywhere inside electronic
devices. This is especially so for various fluids which contain
oil. However, if you are not in a position to de-solder,
dismantle, re-assemble and re-solder a noisy pot, it might not be such a
bad alternative to squirt some pure
isopropyl alcohol into the pot
, give it a few dozen turns, and
leave it for an hour or so to dry out. It is hard for me to
imagine this leading to any difficulties, and there's a good chance
this will fix any noise problems, at least for the coming months or
I have not seen any serious trouble with the many small pots
of the TR-808. However, they can become noisy.
Technology Transplant used to sell replacement sets, so it would be
worth contacting them if you want a whole new set. However I
can't imagine why anyone would need such a set, unless they had been
used far more than the pots in TR-808s I have worked on.
In August 2012 I had success systematically cleaning all these pots in the following manner.
- Squirt in a little isopropyl alcohol into all the pots.
- Turn each pot one or two dozen times.
- Squirt in a little more isopropyl alcohol.
- Use compressed air to blow out the isopropyl alcohol, while turning each pot. This inevitably causes isopropyl alcohol to go elsewhere on the PCB, in addition to where it went after the first two squirts.
- Leave the whole machine to dry for an hour or so before turning it on again.
The section after this one concerns the
Start/Stop and Tap switches. This section concerns all the others.
I have never seen any problems with the three Rotary
switches. It appears
that ALPS still make this range of switches:
however, it could be tricky to find a supplier with the exact same type
for a one-off order.
There is a little push-button switch for Pattern Clear
. They are a
horrible-feeling switch, with no clicking action at all. I
haven't seen one fail. They are probably easy to replace since switches
such as these, with integral buttons, are widely available.
There is a 3 position Basic Variation
toggle switch and a 2 position I/F
A B switch. Technology Transplant sell a pair of
these, newly manufactured:
There is a 3 position slider switch for Pre Scale
. In my experience,
this type of switch can become erratic due to dust, migrating grease
and/or corrosion. My first approach to dealing with this would be
to squirt some isopropyl alcohol into the switch and move it back and
forth a hundred times or so. It is possible to dismantle them for
cleaning, but it is tricky getting them back together, and I am not
sure that this would be superior to leaving it in-situ and putting some
isopropyl alcohol inside it. Likewise the Sync Input / Output
switch on the
I have never worked on the Power
. Some old Roland mains-powered guitar processing
boxes had a similar switch.
The 16 Tact Switches
particular concern. Much of what follows would also be relevant
to the JP-8
(Jupiter 8) which
uses the same switches. If you have whole sets of these failing,
such as groups of four, then you may have a problem with the printed
ribbon cable connecting to the switch board. This is easily fixed
by replacing the ribbon with direct wiring from PCB to PCB.
The TR-808 tact switches are ALPS SKHCABs. These are a stemless
version of the SKHCAA type used in the TB-303 and TR-606. These
are still produced and are known as SKHCBFA010 for the stemless version
(TR-808 and JP-8) and SKHCBEA010 for the TB-303 type with
stems. These are both "1.27 newton" switches, which means about
124 grams as a force.
I do not recommend using these ALPS switches as replacements. Nor
do I recommend using the ALPS sealed tact switches, since they have
what I consider to be a very poor action. Other people find this
acceptable or desirable in the TB-303, but I can't imagine how they
could be better than the Omron switches discussed below. See ../dfish/tact-switches/
measurements and discussion.
I recommend using Omron B3W-4000 sealed tact switches
to replace the ALPS ones
used in the TR-808. These have a similar feel and are sealed
against liquids and dust. So I expect them to last for many years.
The Omron page for these is:
They are available at Farnell, element14 or whatever they call themselves today: http://www.element14.com
. Mouser has them at a better price, but the illustration on their page (2012-03-12) shows another switch. http://www.mouser.com
One approach is to buy a complete replacement switch panel from
These use newly manufactured actuators (black plastic frame)and
buttons. It is not
clear where Roland got their black plastic actuators and coloured
buttons from in the early 1980s. I have never seen them on
non-Roland equipment and they have no ALPS markings. The ALPS
switches have four little slots in the underside of the body which seem
to be for this style of clip-on actuator, so I guess the actuators were
made by ALPS.
Technology Transplant has had
good replicas of the actuators and buttons made for this very purpose. The new LEDs are
somewhat brighter than the originals, without being excessively bright.
However, the buttons are not absolutely
identical in shape or colour, so it
wouldn't look quite right to install one of these buttons in place of
one which went missing on a TR-808.
A common problem is that a button is missing. This is usually not
just the button which has come adrift, but part of the black plastic
actuator as well. Therefore, the entire actuator needs to be
replaced. The actuator contains a LED and snaps around the switch.
The Technology Transplant board is a most welcome development, since it
is the only source of replacement actuators and buttons. However,
they use the same kind of unsealed tact switches as Roland used - so
these switches will fail within a few years, due to dust.
It is possible to significantly prolong the life of these unsealed tact
switches by placing a thin strip of Mylar (technically biaxially-oriented polyester/PET) or other plastic between the
tops of the switches and the bottom of the flexible part of the
actuator. I used Mylar film from 3M "Professional Flip-Frame"
overhead Transparency Protectors (RS7110). The strip is a little
wider (left-right) than the whole set of buttons and, I recall, about
20mm long (front-back).
However, I believe it is better to use Omron sealed tact switches
instead of the new switches in these Technology Transplant switch
Here are some photos of the original ALPS tact switches, the actuators
and the buttons.
The actuator has dual thin plastic hinges between the moving part
which attaches to the button and the fixed part. This can fatigue
with excessive force, or perhaps just lots of use. When this
happens, this part of the actuator and the button will come loose and
fall off the machine. Please be careful if you remove the button,
since the force could break the hinge.
There's no way of repairing
this broken actuator, so the Technology Transplant replacement boards
are definitely required. However, if someone installs a complete
new board, this means they have an old board with at least some good
actuators in them. I have one such board and would be interested
in acquiring one or two more, so I can use the original actuators and buttons to
replace occasional broken actuators on machines I am working on.
This photo shows that the actuator has four little hooks by which it is
aligned with the top surface of the PCB by clipping under the ALPS tact
switch, on the left. There are four slots molded into the switch
body for this purpose. In August 2012, the machine I worked on
had no hooks at the rear of the actuator body. The hooks in the
front (lower on the above photo) are minimal. The above photo
gives the appearance of hooks at the back, but I am not convinced that
these really exist as functional hooks - I suspect they are molding
marks. Perhaps the design was changed. There's no strong
reason to have hooks at the back, since the actuator is held in place
by the LED wires being soldered to the PCB.
The Omron switch on the right has no such slots. Therefore, we
need to file away at these four bottom corners of the Omron switch
body. In August 2012 I decided that we really only need two
slots, on the end of the switch which will be at the front (lower on
the following photo).
The Omron switch bodies are about 0.1mm higher off the PCB (downwards
in the above photo) than the ALPS switches. Since we rely on
these clips at the front of the switch (bottom of the above photo) to
hold the actuator frame flush with the PCB, it is quite tricky filing
the right amount of material off the switches. Too little and the
clip does not go in at all. Too much and the clip will not hold
against the switch body as intended. For doing this faster I used
a small milling bit from a Dremel tool, affixed to a drill press, with
a guide below. This is tricky, so for a single set of switches I
suggest you just use a file. It may help to cut down or file down
the two locating lugs in the middle of the switch, so you can get a
more shallow angle cut when filing.
Installing Omron sealed tact switches is quite a laborious and tricky task
It is worth it, I believe, since these switches feel good and will
probably last for decades.
It is possible to install new LEDs
while doing the above operation. Modern, more efficient, LEDs may
be excessively bright. This can be fixed by increasing the value
of R60, R62, R64 and R66 on the main board.
Also, it is very tricky to bend the LEDs to the exactly correct dimensions
and insert it into the actuator. The existing LEDs have fine
leads and are held in place by heat-deformed plastic. It is a lot of work
to remove the LEDs,
prepare the actuators for new LEDs, bend the new LEDs' leads correctly
and insert them in the actuators, so when soldered they are all at the
correct height. I think few modern LEDs have fine enough leads to
make this a straightforward task. Generally this would be done by
removing the buttons to give access to the LED with its pre-bend leads
to the holes for these leads. However, it is desirable to do this
work without removing the buttons, so as not to place any extra stress
on the plastic hinges of the actuator. I did figure out a way of
doing this, but it is awkward. The new LED can be put through the
hole of the button, with its leads going straight down through a slot
there, not in the path of the old LEDs leads. Then they leads can
be bent sideways to meet the PCB (this may involve using a Dremel etc.
tool to remove some plastic from the base of the actuator).
However . . . the new LEDs need to be narrow enough to fit, and they
are probably not the right height. So the new LEDs must have one or two
small pieces of tubing put around the leads, to lift the base of the
LED body from the top of the slot in the actuator, while allowing the
downwards force on the leads and so the LED body to press the actuator
down to the PCB. The LED leads and body play a crucial role in
holding the rear end of the actuator to the PCB.
A major advantage of the Technology Transplant panels is that they come
with new LEDs which are moderately brighter than the old ones.
My inclination is not to replace these LEDs. It is a lot of work
and I think the old LEDs work OK.
Installing LEDs in multiple colours would be tricky, especially due to the
fact that they are driven in groups (see the schematic for details)
which means it would be difficult to adjust for different LED
efficiencies except by installing individual resistors in series with
Start / Stop and Tap switches
The above information provides what I
consider to be reasonable ways of coping with failure of most of the
TR-808's pots and switches which are likely to fail.
However, it is a work-in-progress figuring
out how to deal with the Start / Stop and Tap switches. The first
subsection here concerns the original switches and how they might
be refurbished. In a further subsection below that I consider alternative switches.
All the switches I have had to replace have been of the type shown
immediately below, with a white stem and two melted lugs to hold the
switch together. One problem refurbishing these is how to get the
switch back together.
Another type has a black shaft and side clips to hold the switch
together. If one of these is in a TR-808 (I have not seen one so)
then refurbishing these is a lot easier because the switches can be
snapped apart and back together with ease. If you are fixing a
switch of this latter type, please read the next section on cleaning
the contacts, but it will be obvious how to put your switch back
Refurbishing the existing switches - melted lug sealing
Here are some photo of original switches. It is made by ALPS but I
don't know its part number. I don't know of any replacement
buttons either, but generally these are not damaged, destroyed or lost.
It is possible that later TR-808s had a different type of ALPS switch
with the same dimensions, but a different form of construction, as
described below. For simplicity I will refer to the above type as
the "original switches".
The key point about them is that the top of the body is affixed to the
bottom part, the bottom 3.5mm, by two heat-melted cylindrical
lugs. We need to "drill these out" to dismantle the switch,
though in the past I recall I cut the switch apart with a knife, which was a bad idea.
I used a small Dremel tool to cut away the ends of the melted
lugs. It doesn't take much, just a fraction of a millimetre,
before the switch can be prised apart with a knife-blade in the
split. By cutting a circular path following where the lug has
been melted sideways a little, it is possible to leave the centre of
the lug intact, but still cut away enough of its widened end to make it
possible to prise the switch apart.
The failure mechanism is the lack of conductivity when the conductive
rubber presses against the metal contacts. One of these switches
was 127k ohms when pressed. It needs to be about 10k or less to
activate the circuit.
My usual approach to dealing with rubber contacts is:
- Use some light, clean, fresh, emery paper (dark-grey paper-backed
abrasive sheet) to abrade away the surface of the conductive
rubber. (I don't put any solvents on this, because once when I
did, the rubber became less conductive.) 240 or 400 grit is
probably a good choice.
- Use isopropyl alcohol to clean the metal contacts.
use a very light touch of fine, fresh, emery paper to roughen the
surface of the contact points just a little. 800 grit is probably
a good choice.
In this case, point 3 is a must. The contacts are selectively
gold plated and examination with a stereo zoom microscope revealed a
dark shiny patina on the gold. In both these switches, one
contact was darker than the other, so perhaps there was some kind of
electrical process affecting the deposition of material. In the
one switch on which I tried acetone, before any abrasion, the acetone
appeared to take off the lighter coating on one contact while not
affecting the darker coating on the other
Scratching with a needle cuts through this to the gold and copper
below. The 240 grit emery paper scratched up the surface
revealing (under the microscope at 30x) some copper rather than gold.
This is not ideal, since the copper is subject to corrosion.
Still, it is better than retaining whatever coating had built up on the
gold surface. When I used 800 grit emery paper, very
lightly, under the stereo zoom microscope, I was able to get a gold-looking set of scratches, without obvious copper colour.
An additional step which is possible with these switches is to rotate
the rubber cup so that fresh material is exposed to the raised
contacts. However, once the conductive rubber is sanded, it is
not obvious where it was pressing into the contacts.
Using these techniques I was able to make one of these switches go
consistently below 1k ohms - 800 to 900 ohms. However, the other
was 2k to 3k and went higher, beyond 10k, as I attempted improvements,
such as smoothing down the abraded metal contacts. I assume the
rubber contact material has degenerated in some way, perhaps due to
some solvent which might have been on contact with it. However,
perhaps my abrasion was too rough.
There are some special paints which are intended to fix problems with
these conductive rubber switches. I haven't used any and I don't
at present know which ones are well-regarded, but such a product might
save the second switch. A quick search found this product: http://www.keypadrepair.com
Also, it would be possible to modify the circuit so it was acceptable to have a 10k or higher resistance.
Assuming the switch contacts can be made to work acceptably, this still
leaves a problem with how to fix the switch back together. One
approach is to use a soldering iron to melt the lugs again. This
would probably work, but next time the switch needs to be refurbished,
it will be difficult to do this again.
In the past I tried epoxying these switches together, but that was
after a careless disassembly which I think involved cutting around the
seam of the switch. I buttressed the sides with epoxy with a
filler, going down to the PCB, but still the switches fell apart after
a little use.
Here is a method I devised in March 2012 which I think will work well.
I would solder the twisted joints and probably give them a bit more of a twist to tighten them up further.
This looks a bit rough, but I think it will be strong enough for
long-term use. It is easy to pull it apart in the future.
Although I filed slots in the bottom, the wire is below the base of the
switch so some care or insulation tape or the like will be required
when mounting the switch in the TR-808.
Replacement switch from an MC-4
I was lucky enough to obtain a few of these switches, which have never
been used used, as spare-parts for the keyswitch panel of an MC-4
micro-composer. These are physically compatible with the switches
above, and perhaps they are used on some TR-808s.
They operate on the same principles, so refurbishing the contacts would be along the lines described above.
This one had a resistance of about 500 ohms.
Both types of switch are intended to snap into a punched steel plate
for the purposes of making a numeric, QUERTY or other similar kind of
"keyboard" set of switches.
The sides unclip easily to reveal:
I believe these are a later model, superior, switch to the melted lug
version. The melted lug version used deformation of the silicone
rubber cup to take up the slack of the button moving down past the
point where the contacts closed. This one uses a coil spring, and
I believe this places less stress on the silicone rubber cup.
Does anyone know a source of these switches or anything like them?
I don't know of any alternative switches. In a July 2011 discussion:
this switch http://www.smallbearelec.com/Detail.bok?no=36
was suggested, but it only has two pins and is apparently a different height.
It can be safely assumed that the people at Technology Transplant have
a better idea of what is manufactured, and can be manufactured, than
most folks. Yet they have no such switches. I think this is
strong evidence that no such switches are currently being manufactured.
Does anyone have suggestions?
These switches measure 13.5 x 13.5mm in the main body - this is the size of hole they are intended to be mounted in.
The four pins are spaced identically to the ALPS and Omron
switches. I was tempted to think of some adaptation of the
existing body and slider mechanism with an Omron tact switch replacing
the current base of the switch. Another idea is to keep the
existing switch body and base, but have a hole through the bottom of it
and the PCB, with a shaft driving an Omron switch mounted below the PCB.